.comment-link {margin-left:.6em;}

Friday, June 10, 2005


Scored A Copy of Zertz, Abstract Rant

I finally scored the last game in the Gipf series that is currently in print. Zertz is a step above Dvonn in my estimation. It is on a par with Yinsh. Gipf remains my favorite of the series. When Tamsk is re-released this summer I will be all over it.

As two-player abstracts go, these Gipf games are all very good. Kris Burm (he designed all of them) did a heck of a job creating simple, elegant, deep, yet-not-too-deep, strategy games.

Is it just me, or have Go and the Gipf games replaced Chess and Checkers as the prime examples of abstract games? Seems like every time I read a review, or a geeklist and the writer needs an example of an abstract game he references Go and Gipf. Just a few short years ago Chess and to a lesser extent Checkers and Chinese Checkers were the go-to games when an example was needed.

I wonder if the Gipf games wouldn't appeal to the mass market. I think they are simple enough to have wide appeal. Chinese Checkers and Mancala sets are available in the box stores, as are Chess sets of various quality. I think the public is hungry for intellectually stimulating abstract games. I think they would have a much better chance of selling well in "Toys R Us" than Settlers of Catan. Plus, the quality components look good setting on a coffee table (Gipf itself excepted).

On a related note; Is it just me, or has Go been getting more and more popular in recent years? Several decades ago the game started creeping out of the Orient, where it has been played for thousands of years. The first western, written description of the game was first published barely over a century ago. Go is played on a 19x19 grid. Each player plays either black or white stones onto the board. The goal is to encompass more territory by your stones than your opponent.

As computers get more and more powerful, Chess programs exist that can only be beaten by the most accomplished Chess masters. Go, in all its simplicity, has defied computer programmers. The mathematical possibilities of Go are many times greater than the mathematical possibilities of Chess, even though pieces are not moved once placed onto the board. Currently, even a new player, with limited understanding of game subtleties can beat a computer program.

I attribute the increase in popularity to increased exposure on the web and on Boardgamegeek in particular. There are a plethora of gamers who loudly proclaim their love for the game, clubs oriented around the game, even professional players.

Personally, I have no desire to study a game for years just to rise to the rank of amateur. I have no desire to study any game for years. The few times I have played Go I was very underwhelmed. I can appreciate the fact that it is purely a cerebral game, even more cerebral (and much older) than Chess, but I can do without it.

(Never heard of Go? Japan and Korea each have a television channel dedicated to Go and I think China might have one as well. This news story gives brief mention to one of the channels) http://www.economist.com/diversions/displayStory.cfm?story_id=3445214

It is true that computers haven't been good Go players, but it's not clear to me that this solely because it's a much more difficult problem. Another factor is just that it seems to me there have been a lot more smart people working on Chess than Go (and even then, it took a long time). Chess had a high profile during the Cold War what with matches actually being televised. With that all behind us, perhaps Go will become the next hard problem programmers turn to in their spare time, and once they do so in earnest, it seems that humans will have as much difficulty retaining their dominance in the game.

I think it's clear that if GIPF ever caught anyone's attention, computers would become very good very fast.
The interesting details about programming Go are 1) the search and state space are so much larger than for chess, and 2) there's no good evaluation function (unlike counting the point values of remaining pieces in chess). The combination of these details means that a new paradigm is needed; the alpha-beta search that "solved" checkers and chess isn't enough for Go on current hardware. The Gnu Go website has some interesting links on coding an AI for Go.
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?